"Describe the face in words" — and the model draws a different person every time. But what if you already have a face — your own, a friend’s, or a brand mascot? Upload a photo: the system reads its features off it and turns it into a full-fledged character.

Upload a photo, get a character

The photo runs through a vision pass: the system fills in the appearance and behavior anchors on its own, checks the content (an NSFW-and-minors gate) and runs a celebrity check — then shows you an editable preview where you can fix anything. After that, optional cartoon stylization (Nano Banana 2, $0.08), and your character is ready.

📷 PhotouploadVision anchorsappearance + gateStylizationopt., $0.08Elementready characterScenesone face

From a photo to an Element: features are captured automatically, stylization is optional, and from there it’s the same consistency pipeline.

In the Image node, the source is Generation / Library / Upload. "Upload" is exactly the entry point for your photo; from there the character travels through the same multi-reference.
In the Image node, the source is Generation / Library / Upload. "Upload" is exactly the entry point for your photo; from there the character travels through the same multi-reference.
Safe by default. The vision pass blocks NSFW and minors and warns about resemblance to celebrities — you work only with your own content and see everything before generation.

From here on — a regular Element

The character you get is the very same Element: it holds its face across every scene through multi-reference and carries over from project to project. The only difference is the start: not a text description, but your photo.

The result — a consistent character across scenes. From a photo or from text, the pipeline downstream is the same.
The result — a consistent character across scenes. From a photo or from text, the pipeline downstream is the same.
  1. Upload a photo. Your own, a friend’s (with consent), or a brand mascot.
  2. Check the features. The system fills in the appearance/behavior anchors — you edit them in the preview.
  3. Stylize if you like. One click turns the photo into a cartoon character (Nano Banana 2).
  4. Use it as an Element. One character — every scene and every future clip.
1 photo→ a ready character
$0.08optional stylization
autofeatures and safety gate
= Elementthe same consistency pipeline
Not "describe the face," but "show the face." The system handles the rest.— Path B, "Soul from a photo"

Build your own hero in Vini Studio

A cloud AI video generator: from an idea in words to a finished clip with sound, 9:16 or 16:9. The project is in closed beta — I am looking for authors and partners.

Get started in the studio →Sign in with Telegram or Google. No account needed in advance — it is created for you.